An optimum data fusion algorithm for distributed detection system

Chair and Varshney (1986) derived an optimal rule for fusing decisions based on the Bayesian criterion. To implement the rule, the probability of detection and the probability of false alarm for each sensor must be known, but this information is not always available in practice. This paper discusses the data fusion algorithms for distributed detection systems when the priory probability is not known. In the condition, if an incorrect priory probability is chosen, the great risk is obtained, but the question is how one can we it? A new data fusion algorithm, a min-max rule in that condition, is presented. This rule considers the worst priory probability, although the risk is more than the minimal average risk under the Bayes rule; this way it has a minimal average risk all in all, when the priory probability is not known. Finally, it is simulated and verified.